`gc+task+time=debug` emits logs like the following:

```
[0.025s][info][gc] Using Parallel
[3.315s][info][gc] GC(0) Pause Young (Allocation Failure) 100M->20M(383M) 61.912ms
[3.315s][debug][gc,task,time] GC(0) VM-Thread 3252631209 3314611442 3315324413
[3.834s][info ][gc ] GC(1) Pause Young (Allocation Failure) 120M->45M(383M) 108.125ms
[3.835s][debug][gc,task,time] GC(1) VM-Thread 3725998296 3833913870 3834515706
[4.261s][info ][gc ] GC(2) Pause Young (Allocation Failure) 145M->68M(383M) 139.535ms
```
The three metrics are #nanoseconds since VM start when a GC cycle starts, reaches the mid-point, ends, respectively. Such metrics are not super useful and can be dropped.